Long Short-Term Memory-Convolution Neural Network Based Hybrid Deep Learning Approach for Power Quality Events Classification

Abstract

This paper presents novel method of deep learning for automatic selection of features and classification of power quality events. The power quality events analysis based on hybrid approach of long short-term memory-convolution neural network is proposed here to ensure reliability, security of supplied power. It is necessary to recognize and classify the power quality events. Power quality disturbance classification methods mostly based on unique feature extraction such as statistical behavior, spatial-temporal information, nonlinear, and nonstationary characteristics of power quality disturbances. The performance of the presented long short-term memory-convolution neural network is evaluated on a set of synthetic power quality disturbances. The developed model is shown to be novel for feature training and classification of various disturbances. Totally different from conventional methods such as Support vector machine (SVM), fuzzy logic, probabilistic neural networks (PNN), and artificial neural networks (ANN). The achieved results show that deep learning based classifier is more efficient than the earlier state of art. In addition, the proposed method can be efficiently used to classify the complex power quality events.